penner/lgl2 is required for the integrity of the photoreceptor layer in the zebrafish retina [RESEARCH ARTICLE]
DISCUSSION
In vertebrates, proper vision depends on the correct specification and differentiation of several neuronal cell types, the formation of synaptic contacts between them and their organization into a highly stratified retina (Amini et al., 2017; Baier, 2013; Morgan and Wong, 1995). The visual process is initiated by PRCs, rods and cones, which capture photons of light by photosensitive pigments. The activation of cone and rod visual pigments triggers the phototransduction cascade, which ultimately transmits the signal into the brain. PRCs are highly polarized cells, forming an apical, light-sensitive organelle, called the OS, and the basal ribbon synapse, which connects the PRCs to second order neurons (horizontal and bipolar cells). To develop these features, PRCs have to establish and maintain apico-basal polarity and to form adhesive contacts, a prerequisite for layer formation. Both processes are tightly coupled and loss of either is linked to the breakdown of the layered structure, retinal degeneration and ultimately blindness. In human, several retinopathies such as retinitis pigmentosa (RP), Leber’s congenital amaurosis (LCA) or Usher syndrome are associated with loss of function of genes regulating polarity or adhesion [reviewed in (Chacon-Camacho and Zenteno, 2015; El-Amraoui and Petit, 2010; Verbakel et al., 2018; Yan and Liu, 2010)]. Therefore, the establishment and maintenance of apico-basal polarity and formation of cellular junctions is crucial for a functional retina. Here we show that knockdown of both maternal and zygotic Lgl2, a protein known to regulate apico-basal polarity in various epithelia, results in the disorganization of the PRC layer. These findings add to our knowledge of the molecular mechanisms regulating morphogenesis of PRCs and may help in understanding their dysregulation in disease.
The retina of pen/lgl2 homozygous mutant zebrafish larvae derived from heterozygous animals laminates and differentiates normally, demonstrating that maternal Lgl2 is sufficient for the development of the embryonic and larval retina. This is different from its requirement in the epidermis, where already the loss of zygotic gene function results in severe defects due to impaired hemidesmosome formation, followed by loss of cell-matrix contacts in the basal layer of the epidermis, blistering and hyperproliferation (Sonawane et al., 2005, 2009). Since the complete MO-induced knockdown (KD) of pen/lgl2 causes severe overall phenotypes (Tay et al., 2013), we injected lower MO concentrations into the pen/lgl2 background. This leads to abnormal organization of the PRC layer in the retina of hetero- and homozygous mutant larvae. From these data we concluded that the maternal contribution of gene expression is sufficient to allow normal retinogenesis, and that the loss of one or two functional copies of pen/lgl2 provides a sensitized genetic background, which allows study into the role of pen/lgl2 in retinal development.
Lgl proteins are well known for their function in establishment and maintenance of apico-basal polarity and junctional complexes in epithelial cells (Cao et al., 2015; Grifoni et al., 2013). Loss of Lgl function has tissue-specific consequences: in some cell types polarity is strongly affected (Bilder et al., 2000; Klezovitch et al., 2004), while in others Lgl is required for junctional integrity and adhesion (Jossin et al., 2017; Sonawane et al., 2005, 2009; Tay et al., 2013). In vertebrates, the situation is complicated by the fact that the genomes encode two orthologues, lgl1 and lgl2. Zebrafish lgl1, for example, plays a role in apical differentiation in the pseudostratified retinal neuroepithelium. Cells with reduced Lgl1 levels retain junctions and overall epithelial integrity, but develop an enlarged apical membrane. This results in increased Notch signaling and, as a consequence, impaired neurogenesis (Clark et al., 2012). Deletion of mouse Llgl1 specifically in embryonic cortical neural stem cells affects the integrity of the apical junctional complex (AJC). Loss of AJC integrity is likely due to impaired interactions between LLGL1 and N-cadherin, which is followed by mislocalization of N-cadherin. Affected neural stem cells are internalized and form rosette-like structures, while their overall apico-basal polarity is retained. Ultimately, this leads to ectopic formation of neurons at the ventricular surface (Jossin et al., 2017). Overall apico-basal polarity is also not affected in the cells of the zebrafish Kuppfer’s vesicle in Lgl2 morphant embryos (Tay et al., 2013), or in basal epidermal cells in pen/lgl2 mutant fish (Sonawane et al., 2005), but both cell types display defects in cellular adhesion upon the loss of Lgl2.
As shown here, loss of zygotic pen/lgl2 activity does not induce any gross modifications of apico-basal polarity and compartment size in PRCs of the larval retina. The inner and OSs develop normally and are similar in size to those of controls. Similarly, no major defects in PRC morphology could be detected upon additional knockdown of the maternal component of pen/lgl2 gene expression, indicating that overall apico-basal polarity is not affected. However, KD of both maternal and zygotic pen/lgl2 function results in defects in the organization of the PRC layer, most likely due to impaired adhesion. This assumption is corroborated by impaired N-cadherin localization at the OLM and the concomitant loss of actin and ZO-1 upon combined deficiency in maternal and zygotic pen/lgl2 function. This phenotype is strikingly similar to that observed upon Llgl1 knockdown in mouse embryonic cortical neural stem cells (Jossin et al., 2017). Furthermore, the retinal disorganization caused by the lack of Lgl2 resembles the loss of N-cadherin (pacrw95 mutants) in the zebrafish retina, which results in lamination defects without affecting neuronal differentiation (Masai et al., 2003). In Lgl2 morphant Kuppfer’s vesicles, basolateral transport of E-cadherin is abnormal (Tay et al., 2013), and similarly, the loss of pen/lgl2 in the basal epidermis causes defects in hemidesmosome formation due to impaired delivery of integrin alpha 6 (Itga6) (Sonawane et al., 2009). These results demonstrate a role for Lgls in regulating cell adhesion via polarized trafficking, and future experiments should address the role of Lgl2 in regulating the dynamics of adherens junction components in the distal retina.
lgl was originally discovered as tumor suppressor gene (Bilder, 2004; Gateff, 1978) and can regulate spindle orientation and asymmetric cell division (Bell et al., 2015; Betschinger et al., 2003; Carvalho et al., 2015; Wirtz-Peitz et al., 2008; Yasumi et al., 2005). In the wild-type zebrafish retina committed photoreceptor precursor cells are dividing parallel to the tissue layer between 60–72 hpf (Suzuki et al., 2013; Weber et al., 2014). Therefore, loss of PRC layering observed in pen/lgl2 MO-injected fish could be the result of hyperproliferation and/or misoriented mitotic spindles, which would place cells outside the plane of the PRC layer, between the PRC layer and the pigment epithelium. Preliminary data on proliferation in the retina revealed no significant difference in the number of phospho-histone H3 positive mitotic cells at 48–72 hpf in morphant versus control retinas in pen/lgl2 clutches (data not shown). Only occasionally dividing cells were found within disorganized cell clusters, indicating that aberrant cell divisions are unlikely to be the cause for the formation of cell clusters. Finally, no cell divisions were detected in the retina of Lgl2 morphants at stages at which proliferation has ceased in the central retina (5 dpf). This is consistent with observations made in Kuppfer’s vesicle upon KD of pen/lgl2, in which no change in proliferation was detected (Tay et al., 2013), but is in marked contrast to those made in the mouse brain neuroepithelium or in the zebrafish epidermis, where loss of Lgl2 leads to hyperproliferation (Klezovitch et al., 2004; Reischauer et al., 2009). In addition, hyperproliferating pen/lgl2 mutant epidermal cells undergo EMT and acquire migratory potential (Reischauer et al., 2009). In pen/lgl2 morphant retinas, PRCs were occasionally found in more basal layers, e.g. in the INL, but their number was very low. Therefore, we favor the conclusion that the major cause for the retinal phenotype upon loss of Lgl2 is the lack of adhesion in the PRC layer.
The detection of disorganized PRC clusters at 3 dpf upon KD of Lgl2 in pen/lgl2 clutches suggests a function for Lgl2 either in organizing the PRC layer during its lamination or at later stages during the maturation of the photoreceptor cell layer. Preliminary data show that retinas of Lgl2 MO-injected pen/lgl2 clutches appear normal at 48 and 54 hpf, and clear signs of disorganization and cell cluster formation appeared in embryos older than 60 hpf. It has to be pointed out, however, that identifying a disorganized PRC layer prior to 60 hpf is challenging, as it is only at this time that the cell layer appears straight and clearly separated from the INL. Therefore, we cannot exclude that the phenotype emerges already during lamination of the PRC layer. However, nuclei/cells mis-localized above the PRC layer were never observed during early stages of development (48–54 hpf), but only accompanied disorganized PRC clusters starting at 60 hpf. Future analyses, including live imaging of fluorescently labelled PRCs to visualize cell morphological and junctional aberrations, are needed to track the onset of cluster formation.
Why are only groups of cells affected by the KD of pen/lgl2, rather than the entire retina? Two scenarios can explain this conundrum. First, variation in MO concentration within retinal cells could induce variable degrees of pen/lgl2 knockdown. Maternal-zygotic pen/lgl2 mutants should be generated to determine the effects of complete Lgl2 loss on the retina, but this was out of the scope of our study. Alternatively, mutant cell clusters could be caused by a defect in one of the cell types that interact with PRCs. For example, the apical processes of MG cells interact with PRCs and form an integral part of the OLM (Gosens et al., 2008; Wei et al., 2006; Zou et al., 2012). Preliminary data from retinas of embryos expressing the MG-specific reporter GFAP:EGFP (Bernardos and Raymond, 2006), injected with Lgl2 MO show loss of apical MG processes within a phenotypical cluster. In addition, cell bodies of some MG are mis-localized basally into the GC layer (data not shown). Whether these defects in MG cells are the cause or the consequence of PRC cluster formation has to be elucidated in the future. Furthermore, our data show strong expression of Lgl2 in cells of the RPE. Defects in the RPE are frequently associated with PRC abnormalities, including degeneration, since OS maintenance depends on a functional RPE (Kevany and Palczewski, 2010; Sparrow et al., 2010; Strauss, 2005). However, we did not detect any gross changes in RPE cell morphology or ultrastructure. Yet, we cannot exclude a role for pen/lgl2 in the RPE, which would impact on the organization of the PRC layer. In the future, cell-type specific inactivation of pen/lgl2 would help to identify in which cell type this gene is required.
